I am currently configuring MC-LAG between two Ciena 3916 and one Cisco 3560CX. On top of that, I have BGP configured on an SVI that is added as a trunk on the Po interface in the Cisco router.
Ciena (Main) 3916_CE01 Port 1---> Cisco 3560_CPE Gi0/2
Ciena (Bak) 3916_CE02 Port 1---> Cisco 3560_CPE Gi0/4
Everything seems to work fine including the switch over from main->backup when simulating power failure on the main CE and/or uplink fibre break.
The only issue I have (and quite game-breaking) is when the main CE01 revert timer (default 5mins) finishes and takes over from the backup CE02 I can see my SVI going down, which then brings BGP down and I get a 40-45 secs outage. The reason why the SVI goes down is that the interface that was UP/UP and working fine at the time of the revert (Gi0/4), goes down BEFORE the main interface goes up (Gi0/2) for 1ms or less.
I have tried this with multiple OS and different Cisco boxes - I wanted some feedback regarding if this is the intended behavior on MC-LAG/LACP or if the issue is more Layer 8, before going back to Ciena and asking them why their CE drops the up/up interface before bringing up the backup one.
No comments:
Post a Comment